Richard, have you looked at the IDE mess? That "feature" is a bug, the way
it was implemented - and considering that it's implementable at a
different level much more cleanly for the (few) people who actually need
it...
Also note that performance is likely to _increase_ by removing that stupid
feature - using DMA to do the actual IO and them byteswapping in some
higher level than the driver is likely to be a _lot_ faster than doing PIO
(and byteswap in-place, resulting in random mmap corruption).
Do you realize that because the current bswap writeback reverses the bytes
in place, you can actually seriously corrupt your filesystem by just being
unlucky in timing (ie a bswap on a dirty metadata block at the same time
another process accesses it)?
It's a BUG guys, not a feature.
Linus
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/